On Thu, Nov 26, 2009 at 3:44 PM, Allan Odgaard
<[email protected]> wrote:
> On 26 Nov 2009, at 22:29, Peng Yu wrote:
>
>>> [...]
>>
>> This solution seems awkward to me. There maybe not be a good pattern that
>> catch both a and b. Should gnu make have a more natural solution to this
>> problem?
>
> I had the problem with ANTLR which is a parser-generator and generates
> multiple files.
>
> I ended up with a solution a la:
>
> grammar.cpp: grammar.g
> antlr …
> lexer.cpp: grammar.cpp
> touch $@
>
> Here antlr generates both grammar.cpp + lexer.cpp, but as far as make knows,
> lexer.cpp is generated from grammar.cpp which is what depends on grammar.g.
Although this works for a program that generated only a small number
of file, it is still cumbersome for a program that generated thousands
of files, right?
Suppose 'program.exe' generate output1, output2,...,output100000,
The rules will look like.
output1: program.exe
./program.exe
output2: output1
touch $@
...
output100000: output1
touch $@
Also, when I delete file 'b' but not file 'a' after I run make once,
the target 'b' will not be updated correctly.
.PHONY: all
all: a b
a: command.sh
./command.sh
b: a
touch $@
_______________________________________________
Help-make mailing list
[email protected]
http://lists.gnu.org/mailman/listinfo/help-make
